qdrant-demo

This repository contains code for a demo utilising:

• Qdrant
• Streamlit
• arxiv API (python wrapper)

It provides a text similiarity search based on the 1000 most recent publications from arxiv. It uses Qdrant's vector database and streamlit as a frontend layer.

Requirements

Install python requirements:

pip install -r requirements.txt

You also need Docker to run Qdrant.

How to run

Running this app locally requires to prepare the data first.

Pull recent Qdrant image:

docker pull qdrant/qdrant

Now run the service inside Docker:

docker run -d -p 6333:6333 \
    -v $(pwd)/qdrant_storage:/qdrant/storage \
    qdrant/qdrant

After starting the service, upload the data by running:

# Init neural search
python -m init_neural_search

Finally, you can launch the application:

streamlit run main.py

This should expose the application at http://localhost:8501